test: document webpack source map behaviour for column=0 lookups#299
Open
test: document webpack source map behaviour for column=0 lookups#299
Conversation
Add tests for SourceMapper.mappingInfo with a synthetic webpack-style single-line bundle to document the known limitation introduced by #248. Background ---------- PR #81 changed originalPositionFor to always try LEAST_UPPER_BOUND first (then fall back to GREATEST_LOWER_BOUND). This was reverted in #106 because it broke webpack source maps: for real non-zero columns LEAST_UPPER_BOUND finds the *next* mapping (≥ column) rather than the one at the column, returning wrong function names. PR #248 fixed that regression by using LEAST_UPPER_BOUND only when column === 0, and GREATEST_LOWER_BOUND otherwise. This correctly handles Node.js ≥ 25 where V8's LineTick struct carries real column numbers. Residual limitation (Node.js < 25) ----------------------------------- On Node.js < 25, the LineTick struct has no column field. The C++ layer therefore always emits column=0 for every LineTick sample. With column=0, the sourcemapper uses LEAST_UPPER_BOUND, which finds the *first* mapping on the line. In a webpack bundle (all output on one line) every function maps to the same first source function in the map. This is not a regression vs. the pre-#248 state: before #248, those functions were simply unmapped (column=0 + GREATEST_LOWER_BOUND → nothing ≤ 0 → null → falls back to generated name/file). Both outcomes are imperfect; #248 trades "unmapped" for "mapped to first function", which may or may not be preferable depending on the use case. The two new tests pin both behaviours explicitly so any future change to this logic is immediately visible. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Overall package sizeSelf size: 2.04 MB Dependency sizes| name | version | self size | total size | |------|---------|-----------|------------| | source-map | 0.7.6 | 185.63 kB | 185.63 kB | | pprof-format | 2.2.1 | 163.06 kB | 163.06 kB | | node-gyp-build | 3.9.0 | 8.81 kB | 8.81 kB |🤖 This report was automatically generated by heaviest-objects-in-the-universe |
szegedi
changed the title
szegedi
requested review from
IlyasShabi,
nsavoire and
r1viollet
as code owners
BenchmarksBenchmark execution time: 2026-03-13 12:45:12 Comparing candidate commit d6f79f2 in PR branch Found 0 performance improvements and 0 performance regressions! Performance is the same for 91 metrics, 29 unstable metrics. |
When lineNumbers is enabled, the column is always zero. If only one occurence of a call occurs on one line then that is correctly selected. However, in cases where the same function is called multiple times in the same line it will be unable to differentiate them and would use the unmapped value. This now makes it select the first call in the line as a best-guess for the match, and in Node.js v25 will use the new column field in LineTick to select the correct column where possible.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Background
PR #81 changed
originalPositionForto always tryLEAST_UPPER_BOUNDfirst (then fall back toGREATEST_LOWER_BOUND). It was reverted in #106 because it broke webpack source maps: for real non-zero columns,LEAST_UPPER_BOUNDfinds the next mapping (≥ column) rather than the mapping at that column, returning wrong function names.PR #248 fixed that regression by only using
LEAST_UPPER_BOUNDwhencolumn === 0, andGREATEST_LOWER_BOUNDotherwise. This is correct for Node.js ≥ 25, where V8'sLineTickstruct carries real column numbers.There was still a concern that #248 will be incorrect for Webpack, so I decided we should investigate this, and either fix it, or at least understand the behavior, as I'd finally like to add #248 to a release. This is what we ended up doing, understanding the behavior, and codifying it in tests so we can detect future regressions.
Investigation findings
Is the original webpack failure reproduced by #248? — Partially.
GREATEST_LOWER_BOUND→ correct ✓GREATEST_LOWER_BOUND→ correct ✓GREATEST_LOWER_BOUND→ null → unmappedLEAST_UPPER_BOUND→ first mapping → mapped to wrong functionOn Node.js ≥ 25 everything is correct. #248 is a genuine fix for that platform.
Residual limitation on Node.js < 25: The
LineTickstruct has nocolumnfield on those versions, so the C++ layer always emitscolumn=0for every sample. Withcolumn=0,LEAST_UPPER_BOUNDfinds the first mapping on the line. In a webpack bundle (all output minified onto one line) every function gets attributed to whichever source function appears first in the map.This is not a regression vs. the pre-#248 state: before #248 those functions were simply unmapped (falling back to generated names). Both outcomes are imperfect; #248 trades "unmapped" for "mapped to first function". On the balance, this is likely an improvement — a wrong-but-close mapping is more useful in a profile than no mapping at all.
What this PR does
Adds two unit tests to
test-sourcemapper.tsusing a synthetic webpack-style source map (three functions at columns 10, 30, 50 on a single line, from three different source files):column > 0): verifies each function resolves to its correct source.column = 0): documents the known limitation — all three functions resolve to the first mapping on the line. The test assertions encode the current behaviour explicitly so any future change is immediately visible.No behaviour changes; tests only.
🤖 Generated with Claude Code